product iamge

Introduction to Deep Learning for Healthcare

This document was uploaded by one of our users. The uploader already confirmed that they had the permission to publish it. If you are author/publisher or own the copyright of this documents, please report to us by using this DMCA report form.

Download
Search on Amazon

Simply click on the Download Book button.

Yes, Book downloads on Ebookily are 100% Free.

Sometimes the book is free on Amazon As well, so go ahead and hit "Search on Amazon"

This textbook presents deep learning models and their healthcare applications. It focuses on rich health data and deep learning models that can effectively model health data. Healthcare data: Among all healthcare technologies, electronic health records (EHRs) had vast adoption and a significant impact on healthcare delivery in recent years. One crucial benefit of EHRs is to capture all the patient encounters with rich multi-modality data. Healthcare data include both structured and unstructured information. Structured data include various medical codes for diagnoses and procedures, lab results, and medication information. Unstructured data contain 1) clinical notes as text, 2) medical imaging data such as X-rays, echocardiogram, and magnetic resonance imaging (MRI), and 3) time-series data such as the electrocardiogram (ECG) and electroencephalogram (EEG). Beyond the data collected during clinical visits, patient self-generated/reported data start to grow thanks to wearable sensors’ increasing use.  The authors  present deep learning case studies on all data described.

Deep learning models: Neural network models are a class of machine learning methods with a long history. Deep learning models are neural networks of many layers, which can extract multiple levels of features from raw data. Deep learning applied to healthcare is a natural and promising direction with many initial successes. The authors cover deep neural networks, convolutional neural networks, recurrent neural networks, embedding methods, autoencoders, attention models, graph neural networks, memory networks, and generative models. It’s presented with concrete healthcare case studies such as clinical predictive modeling, readmission prediction, phenotyping, x-ray classification, ECG diagnosis, sleep monitoring, automatic diagnosis coding from clinical notes, automatic deidentification, medication recommendation, drug discovery (drug property prediction and molecule generation), and clinical trial matching.

This textbook targets graduate-level students focused on deep learning methods and their healthcare applications. It can be used for the concepts of deep learning and its applications as well. Researchers working in this field will also find this book to be extremely useful and valuable for their research.

Author(s): Cao Xiao, Jimeng Sun
Edition: 1
Publisher: Springer
Year: 2021

Language: English
Pages: 243

Preface
Contents
1 Introduction
1.1 Motivating Applications
1.1.1 Diabetic Retinopathy Detection
1.1.2 Early Detection of Heart Failure
1.1.3 Sleep Analysis
1.1.4 Treatment Recommendation
1.1.5 Clinical Trial Matching
1.1.6 Molecule Property Prediction and Generation
1.2 Who Should Read This Book?
1.3 Who Are the Authors?
1.4 Book Organization
1.5 Exercises
2 Health Data
2.1 The Growth of Electronic Health Records
2.2 Health Data
2.2.1 The Life Cycle of Health Data
2.2.2 Structured Health Data
2.2.3 Unstructured Clinical Notes
2.2.4 Continuous Signals
2.2.5 Medical Imaging Data
2.2.6 Biomedical Data for In Silico Drug Discovery
2.3 Health Data Standards
2.4 Exercises
3 Machine Learning Basics
3.1 Predictive Modeling Pipeline
3.2 Supervised Learning
3.2.1 Logistic Regression
3.2.2 Softmax Regression
3.2.3 Gradient Descent
3.2.4 Stochastic and Minibatch Gradient Descent
3.3 Unsupervised Learning
3.3.1 Principal Component Analysis
3.3.2 Clustering
3.4 Evaluation Metrics
3.4.1 Evaluation Metrics for Regression Tasks
3.4.2 Evaluation Metrics for Classification Tasks
Binary Prediction for Classification
Real-Value Prediction for Classification
Multi-Class Classification
3.4.3 Evaluation Metrics for Clustering Tasks
3.4.4 Evaluation Strategy
3.5 Exercises
4 Deep Neural Networks (DNN)
4.1 A Single Neuron
4.1.1 Activation Function
4.1.2 Loss Function
4.1.3 Train a Single Neuron
4.2 Multilayer Neural Network
4.2.1 Network Representation
4.2.2 Train a Multilayer Neural Network
Forward Computation
Backward Propagation
4.2.3 Parameters and Hyper-Parameters
4.3 Case Study: Readmission Prediction from EHR Data with DNN
4.4 Case Study: DNN for Drug Property Prediction
4.5 Exercises
5 Embedding
5.1 Overview
5.2 Word2Vec
5.2.1 Idea and Formulation of Word2Vec
5.2.2 t-Distributed Stochastic Neighbor Embedding(t-SNE)
5.2.3 Healthcare Application of Word2Vec
5.3 Med2Vec: Two-Level Embedding for EHR
5.3.1 Med2Vec Method
5.4 MiME: Embed Internal Structure
5.4.1 Notations of MIME
5.4.2 Description of MIME
5.4.3 Experiment Results of MIME
5.5 Exercises
6 Convolutional Neural Networks (CNN)
6.1 CNN Intuition
6.2 Architecture of CNN
6.2.1 Convolution Layer: 1D
6.2.2 Convolution Layer: 2D
6.2.3 Pooling Layer
6.2.4 Fully Connected Layer
6.3 Backpropagation Algorithm in CNN*
6.3.1 Forward and Backward Computation for 1D Data
6.3.2 Special CNN Architectures
LeNet
AlexNet
VGG
GoogLeNet Inception Net
ResNet
DenseNet
6.4 Case Study: Diabetic Retinopathy Detection
6.5 Case Study: Skin Cancer Detection
6.6 Case Study: Automated Surveillance of Cranial Images for Acute Neurologic Events
6.7 Case Study: Detection of Lymph Node Metastases from Pathology Images
6.8 Case Study: Cardiologist-Level Arrhythmia Detection and Classification in Ambulatory ECG
6.9 Case Study: COVID X-Ray Image Classification
6.10 Exercises
7 Recurrent Neural Networks (RNN)
7.1 RNN Fundamentals
7.2 Backpropagation Through Time (BPTT) Algorithm
7.2.1 Forward Pass
7.2.2 Backward Pass
7.3 RNN Variants
7.3.1 Long Short-Term Memory (LSTM)
7.3.2 Gated Recurrent Unit (GRU)
7.3.3 Bidirectional RNN
7.3.4 Encoder-Decoder Sequence-to-Sequence Models
7.4 Case Study: Early Detection of Heart Failure
7.5 Case Study: Sequential Clinical Event Prediction
7.6 Case Study: De-identification of Clinical Notes
7.7 Case Study: Learning to Prescribe Treatment Combination for Multimorbidity
7.8 Exercises
8 Autoencoders (AE)
8.1 Overview
8.2 Autoencoders
8.3 Sparse Autoencoders
8.4 Stacked Autoencoders
8.5 Denoising Autoencoders
8.6 Case Study: ``Deep Patient'' via Stacked DenoisingAutoencoders
8.7 Case Study: Learning from Noisy, Sparse, and Irregular Clinical Data
8.8 Exercises
9 Attention Models
9.1 Overview
9.2 Attention Mechanism
9.3 Case Study: Attention Model over Longitudinal EHR
9.4 Case Study: Attention Model over a Medical Ontology
9.5 Case Study: ICD Classification from Clinical Notes
9.6 Case Study: Heart Disease Detection fromElectrocardiography
9.7 Exercises
10 Graph Neural Networks
10.1 Overview
10.2 Notations and Tasks on Graphs
10.2.1 Notations and Operations
10.2.2 Tasks on Graphs
10.3 Graph Neural Networks
10.4 Graph Convolutional Networks
10.5 Message Passing Neural Network (MPNN)
10.6 Graph Attention Networks
10.7 Case Study: Neural Fingerprint in Drug Molecule Embedding with GCN
10.8 Case Study: Decagon Modeling Polypharmacy Side Effects with GCN
10.9 Case Study: Deep Learning Approach to AntibioticDiscovery
10.10 Case Study: STAN Spatio-Temporal Attention Network with GAT for Pandemic Prediction
10.11 Exercises
11 Memory Networks
11.1 Original Memory Networks
11.2 End-to-End Memory Networks
11.3 Self-Attention and Transformer
11.4 BERT: Pre-training of Deep Bidirectional Transformers
11.5 Case Study: Doctor2Vec—Doctor Recommendation for Clinical Trial Recruitment
11.6 Case Study: Medication Recommendation
11.7 Case Study: Pre-training of Graph Augmented Transformers for Medication Recommendation
11.8 Exercises
12 Generative Models
12.1 Generative Adversarial Networks (GAN)
12.1.1 The GAN Framework
12.1.2 The Loss Function of Discriminator
12.1.3 The Loss Function of Generator
12.1.4 Caveats of GAN
12.2 Variational Autoencoders (VAE)
12.2.1 VAE from Deep Learning Perspective
Review of Autoencoder
VAE as Two Neural Networks
12.2.2 VAE from Probabilistic Model Perspective
Justification of the VAE Loss
VAE Based on Variational Approximation
12.2.3 Reparameterization Trick
12.3 Case Study: Generating Patient Records with GAN
12.4 Case Study: Molecule Generation Using VAE
12.5 Case Study: MolGAN an Implicit Generative Model for Small Molecular Graphs
12.6 Exercises
Bibliography